Probability-based service safety life prediction approach of raw and treated turbine blades regarding combined cycle fatigue

To avoid the use of specific conversion coefficients with high expense and unacceptable prediction accuracy, a probability-based prediction method is proposed by considering probabilistic feature parameters, to predict the service safety life (SSL) of aeroengine turbine blades. The direct correlation between laboratory remaining life (LRL) and SSL was firstly established by considering probabilistic feature parameters. By conducting Combined high and low Cycle Fatigue (CCF) tests of turbine blades, the effectiveness of the developed method was validated based on the failure event. The proposed method was further verified by predicting the SSL of treated blades with certain operation time. In respect of the studies, it is illustrated that (1) the SSL of turbine blade can be reasonably reflected by the LRL in respect of probabilistic feature parameters; (2) the prediction errors of the raw and treated blades are 2.2% and 12.7%, respectively, indicating that the developed probability-based prediction method has acceptable prediction precision and is an effective method in the SSL prediction of aeroengine turbine blades; (3) the developed method needs less samples than the specific conversion coefficients method, indicating that the SSL prediction of turbine blade needs fewer time and costs. The efforts of this study provide a promising approach for the SSL prediction of turbine blades, offer a useful guidance for the service life management of aeroengine turbine blades to reduce the cost of expense and time and enhance the safety of aeroengine operation . (c) 2021 Elsevier Masson SAS. All rights reserved.

Automated summary

A probability-based prediction method was proposed to predict the service safety life (SSL) of aeroengine turbine blades, considering probabilistic feature parameters. The method showed acceptable prediction accuracy and required fewer samples compared to the specific conversion coefficients method, indicating lower prediction time and costs for SSL prediction of turbine blades.